The central issue in any Open Source business model is how to convert software that is free on the Internet into revenue that can be booked under US GAAP, hence the term invented by some clever person, "Conversion Model." The issue boils down to how to convert the free stuff developed by volunteers into cash?

Linus Torvald invented the first conversion model. By owning the Linux trademark, he has maintained control of brand and parleyed that control into worldwide celebrity. Celebrities make money in lots of ways - when you become a celebrity they will let you in on their secrets. (References: http://en.wikipedia.org/wiki/Linus_Torvalds) Linus delegates the management of the trademark to Linux International that is led by Jon "maddog" Hall (http://en.wikipedia.org/wiki/Jon_%22maddog%22_Hall).

Ian Murdock, the creator of Debian, - which stands for the combination of Debra, his wife's name, and his name, Ian - chose a different path. Debian is owned by Software in the Public Interest so no one really benefits or converts the trademark into cash (http://lists.spi-inc.org/cgi-bin/admin/spi-trademark).

The Apache Software Foundation couldn't trademark the word Apache - the Native American tribe just won't go along with the idea - so Apache is really open and free.

In the early '90, the concept of a Linux "distribution" was invented. Technically, that is technically in the legal sense, there is nothing in a GNU General Public License preventing anyone from selling free software to willing buyers; you just have to find people who are willing to pay for something that is free. So someone came up with the clever idea of the "media kit." A "media kit" is a box filled with CDs and manuals. It's analogous to how software was sold in the "personal" computer era from 1982-1995 - the halcyon days of Lotus, Microsoft, Sierra Online, and Maxis. The SuSE Linux business unit of Novell is actually the most successful of the "media kit" companies. It still sells SuSE Linux Professional 9.3 for $99.95. Rumor has it that Novell SuSE Linux sells about 250,000 units per year - many to students under a discount program - and generates revenue of $15 million per year from the "media kit" or "box" product line. Mandriva, another remaining practitioner of the "media kit" model, appears to have sold at least 30,000 units of its Mandriva Linux and generated about $3 million in revenue, according to financial information on its Web site (www.mandriva.com/company/investors/newsletter the date of the latest report is a typo and should be May 17, 2005).

However, no one ever made a lot of money converting free software to a "media kit." It's unlikely that SuSE Linux makes a profit on its "box" business (because they never brag about it). Mandriva loses money selling media kits (it makes an operating profit only due to R&D grants). Red Hat gave up on the business model and shifted its focus to the enterprise.

It's not hard to understand why this conversion model can't make money. First, the price point is too low at $100 per unit. You have to sell a huge number of units to achieve critical mass. Second, the cost of goods sold is too high at 20%. For example, Mandriva includes eight CDs, one DVD, two manuals, and one guide in its media kit; that's a lot of paraphernalia for a hundred bucks.

Red Hat is the company that figured out a way to evolve beyond the "media kit." It invented the idea of the "enterprise server." The key innovation of the "enterprise server" is that the price point is much higher, from $349 to $2,499. The second innovation is that it's a subscription so you get to pay and pay and pay - that's right, every year cough it up or you're cut off.

With an enterprise server, the customer gets access to a maintenance portal - sometimes called a maintenance network (though I have never understood the logic of a network) or maintenance Web.

The enterprise conversion model has been good to Red Hat. By taking free software and certifying that it works and promising to provide maintenance, training, and support, Red Hat's enterprise server subscription revenue has increased from nothing in 1999 to an annual run rate of about $200M in 2006 (see Figure 1). During the same period, its media kit business has gone from $10M in 1999 to a peak of $20M in 2001 to nothing today.

Table 1 shows how the Red Hat business model has changed over time since going public. In 1999, the company presented itself to Wall Street as a new economy software company that sold Web advertising and services. The software moniker was dropped in 2000 in favor of the "subscription" model. In early 2001, Web advertising was abandoned. During the middle of 2001, the company was positioning itself as a services player and was highlighting its networking consulting, embedded development, and Open Source services businesses. In 2002, it collapsed its networking consulting and Open Source services businesses into enterprise services. Between the middle of 2003 and the end of 2004, it broke out its retail or "media kit" business from its enterprise server to highlight the strong growth of the enterprise segment. So by the first quarter of 2005, its business model has matured to the point that it only has two segments: enterprise and embedded. I predict that the embedded segment will disappear in the next few quarters.

During its search for a viable business model, Red Hat's stock has fluctuated from its "bubble" high of almost $150 - what were we thinking - to a post-9/11 low of $3 - what were we thinking. After 9/11 Red Hat had become a "Budweiser" stock, i.e., you would have done better to spend the money on beer, drank the beer, and turned in the cans for the deposit refund rather than buying the stock - and it would have felt much better. The stock has recovered since 2001 and is now trading at $15.25 per share (see Figure 2). The stock is still pricey at 61x trailing 12 months earnings - what were they thinking in 2000. But the business is solid, profitable, and growing.

There is a lot more to say about various versions of the conversion model. I'll share some information on JBoss, MySQL, WIN4LIN, and OPEN-XCHANGE in coming articles.

Open Source Business Models
√ Donation
√ Advertising
√ Conversion
- Brand Ownership
- Media Kit
- Maintenance
- Support
- Add-ons (Dual License)
√ Professional Services